When inflated beliefs about future value come into conflict with actual underlying value, conditions are right to create a bubble. Housing prices will always go up! Having a dot.com in your name will always increase value! If you can get to scale, you’ll win the digital battle! If you don’t invest in Crypto now, you’ll miss a once-in-a-lifetime opportunity! Let’s consider the evidence with respect to targeted Internet advertising.
Inflection points always come along gradually, then suddenly
In my book “Seeing Around Corners,” I make the point that inflection points, those events that change reality by a factor of 10, are often burbling along for ages, sometimes for decades. Then, when they pass a tipping point, the general reaction is “oh, this came completely out of the blue!” Take, for example, the COVID-19 pandemic.
People, just because you’re not paying attention doesn’t mean the signals aren’t there. Consider this Wall Street Journal article from 2017. Called “Predicting the Next Pandemic,” its first line is literally “Where will the next pandemic come from? Likely from bats.”
Which brings me to the subject of Internet advertising. The kind of advertising that uses personal information about you, freely given or sneakily extracted to try to sell you stuff. Specifically, the programmatic ads that use data to figure out which ads to show you, on the assumption that this will influence your decisions. To give you a sense of how important this kind of advertising is to many of our most significant companies, according to an observer, “97% of Meta’s and 80% of Google’s total revenue comes from digital advertising – the majority of which is programmatic.” Amazon has a huge footprint in advertising as well – according to Statista, it raked in close to $47 billion in ads. The total numbers are staggering – again, according to Statista, worldwide spending on digital advertising will be about $740.3 billion in 2024.
What if all that spending were based on misconceptions about personal behavior and the illusion of precision offered by the platforms that host the ads?
This is the central argument in a (mostly ignored) 2020 book by Tim Hwang, The Sub-Prime Attention Crisis. A blurb for the book summarizes his argument, likening what’s going on to the period leading up to the financial crisis of 2008. As it says, “From the unreliability of advertising numbers and the unregulated automation of advertising bidding wars, to the simple fact that online ads mostly fail to work, Hwang demonstrates that while consumers’ attention has never been more prized, the true value of that attention itself—much like subprime mortgages—is wildly misrepresented.”
In the period leading up to 2007, programmatic trading sunk huge sums of money into financial instruments that basically turned out to be worthless, precipitating the global financial meltdown we now call the Great Recession. My personal favorite take on explaining what happened is the book by Michael Lewis and the subsequent 2015 movie “The Big Short.”
The movie, by the way, illustrates one of the biggest problems with seeing around corners – while you can be right on what will happen, you’re almost certainly going to be wrong about the timing. For instance, economist Paul Krugman started warning about a housing bubble at least as early as 2005. Nothing happened. 2006? Nothing happened. It wasn’t until mid-2007 that the news broke, and some people began to praise his earlier thinking, after many an eye-roll in the earlier period.
Evidence of a bubble
Investopedia tells us that a bubble “refers to a situation where the price for something—an individual stock, a financial asset, or even an entire sector, market, or asset class—exceeds its fundamental value by a large margin.” So, what is the evidence that the $740 billion advertisers are sinking into targeted digital ads may not be justified by the effectiveness of those ads?
50% of spending goes to middlemen and much of it is untraceable
From the perspective of an ad buyer, the more funds that go into creating compelling content and targeting just the right potential buyers, the better. Everything else is waste. In a 2020 study by PwC, researchers found that roughly half of a brand’s digital marketing spend was “absorbed” (corporate-speak for ‘charged by the ad tech company’) before publication. Further, almost a third of the costs of placing ads were “completely untraceable.” To give an idea of the order of magnitude, the researchers were conducting a so-called end-to-end analysis of the ad spending programs. Of 267 million ads in their sample, it was only possible to match the end-to-end process for 31 million of them, only about 11%. It boggles my mind that companies who can measure the costs of things like fuel consumption to the fraction of a penny are happy enough to chuck money at a system without, seemingly, any accountability.
A magical belief in the power of micro-targeting
The whole programmatic ad concept is based on what seems like an obvious idea – if I know what someone is interested in, I can send them ads that are likely to appeal to those interests. And yet, when put to the empirical test, this obvious idea is not borne out. In a stunning conclusion from a 2019 study, the authors find that “When investigating gender (being male) and age (three different tiers: 18–24, 25–34, and 35–44 years) individually, we find that digital audiences for gender are, on average, less often correct than random guessing (accuracy of 42.3%).”
Pervasive ad-blocking
Obviously, a prospective customer can’t be moved to buy something based on an ad they never see. Intense dislike of the way most ads are served up has resulted in a huge business selling technology that can block pesky ads. One source says that ad-blocking represents a $54 billion loss in 2024, in the sense that ads were paid for that intended users never saw. Incidentally, if advertisers were sending us information that is useful, about things we are interested in, we would all be a lot less likely to block ads, wouldn’t we?
Digital ad fraud
This refers to the artificial creation of the appearance of clicks and interactions which simply extract revenue from the advertiser without actually involving a human buyer. An article by Gilad Edelman in Wired cites Hwang quoting a 2017 study that shows that as much as 56% of all display ad dollars were lost to fraudulent or unviewable content. This problem of course is exacerbated by the lack of opacity of the operations of the big digital ad exchanges.
Lack of effectiveness, poor end placement and “whitelisting”
JP Morgan Chase conducted a careful study in 2017 of where their ads were appearing on 400,000 websites. When the bank looked into it, only 12,000 of those ads led to any activity (like a click) other than an impression. Of those, 7,000 turned out to be sites the company didn’t want to advertise on, leaving them with 5,000 sites they approved. Programmatic placement showed that it has little utility when sales didn’t budge, even after the bank cut way back on its budget for those kinds of ads. Similarly, the New York Times cut off all programmatic ad buying in Europe after the implementation of the General Data Protection Regulation (GPDR). What happened to sales? Absolutely nothing, again suggesting the ads didn’t do much.
Selection bias
Another huge problem with Internet ads is something academics call “selection bias.” The “selection effect” occurs when people do see your ad but were already going to take the call to action – click, buy, register, or download – you wanted them to, meaning your advertising spending was wasted. This is opposed to the advertising effect, in which people see your ad and that’s why they followed the call to action.
Berkeley economics Professor Steve Tadelis got the opportunity to conduct a real-world experiment on this due to a spat between eBay and the MSN network. eBay wanted to negotiate lower prices and suspended their advertising against the keyword “eBay.” Tadelis and his team analyzed the effects of the ad stop. The results were stunning – the same amount of traffic that had come from paid links was now coming through ordinary search links. Basically, eBay was wasting $20 million a year on ads targeting the keyword “eBay.”
Tadelis conducted further studies on the impact of stopping key word ads for other product categories. The conclusion? As an article reports, “For every dollar eBay spent on search advertising, they lost roughly 63 cents, according to Tadelis’s calculations. The experiment ended up showing that, for years, eBay had been spending millions of dollars on fruitless online advertising excess, and that the joke had been entirely on the company.”
Back to our classic definition of a bubble – when the price of something exceeds its fundamental value. When looked at through that lens, programmatic ads look bubbly indeed.
If programmatic advertising is so flawed, why do companies keep investing in it?
Justin M. Rao, at the time with Microsoft Research came to an interesting conclusion with respect to deeply held, evidence-free beliefs by studying a group of people associated with the 2011 “End of the World” prediction. This was that on a specific day, all of humanity would ascend to heaven. He and his research team offered prizes to cult members of $5.00 “today” versus increasing amounts of money (up to $500) after the special date. He found that people who sincerely believed in the prophecy would gladly accept the funds today but wouldn’t take any amount of money offered after the special date.
“Beliefs formed on insufficient evidence seem tough to move,” he wrote.
Jessie Frederick and Maurits Martijn note, “It might sound crazy, but companies are not equipped to assess whether their ad spending actually makes money. It is in the best interest of a firm like eBay to know whether its campaigns are profitable, but not so for eBay’s marketing department.
Its own interest is in securing the largest possible budget, which is much easier if you can demonstrate that what you do actually works. Within the marketing department, TV, print and digital compete with each other to show who’s more important, a dynamic that hardly promotes honest reporting.”
This is eerily reflective of conversations happening back in the day of the housing bubble. Perhaps one of the most famous was uttered by then-Citibank’s CEO Chuck Prince in 2007. “When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing,”
Well, dance away, internet advertisers. But by the time this bubble does what bubbles always do, you may well regret not putting those empty dollars into something that could make a bigger contribution to your future – such as innovation or transformational change.